Skip to content
focoit.com
02My own practice, since 2026

AI governance, enforced in code.

I run AI coding agents under one standard across my own delivery repositories: the policies, specs and gates that decide how an agent plans, writes and reviews code. A policy no gate enforces is a suggestion, so the rules that matter are checks that run at every commit.

policy areas
5
repositories under it
10
models in the final gate
0
in use since
Jul 2026

The rules

What runs at every commit.

  1. 01In the code projects, every non-trivial change starts from a backlog item and a spec, approved before code is written, and a spec that fails strict validation blocks the commit.
  2. 02The agent that writes the code and the agent that reviews it come from different model families; reviewers report, they do not fix.
  3. 03The last word is a deterministic script, in the pre-commit hook and in CI. It is never a model.
  4. 04Secrets are blocked at the commit, and client code is never routed to a hosted free service.
  5. 05What an agent reads is data, not instructions; its tools are granted by an allowlist.
  6. 06A policy change propagates to the adopting repositories without overwriting what each one adapted.
  7. 07Ten repositories run under it as of November 2026, two of them on hold.
For technical evaluators: show the stack

Platform

Markdown policies · a PowerShell gate · OpenSpec · git hooks · GitHub Actions · gitleaks · Claude Code and other vendors' agents.

Five policy areas

Five policy areas
AreaWhat it doesWhat it preventsWhere it is enforced
Model execution and routingPlanner, executor and reviewer are separate seats; the reviewer is never from the executor's model familyA model approving its own work; client code reaching a hosted free serviceThe task loop and its policy
Secret scanningScans every staged change for keys and tokensA credential reaching the repositoryPre-commit hook and CI: blocks
Bounded reviewer agentAn AI first reviewer with a narrow mandateAn agent writing business code or changing the gatePull request
Spec-first delivery (code projects)A backlog item and an approved spec come before code; specs are validated strictlyUnspecified or malformed work entering the loopPre-commit hook: a failing spec blocks; the "no spec" reminder advises
Policy propagationA registry of adopting repositories; mirrored documents are updated, project-owned ones are notDrift between repositories; a central update clobbering local decisionsA propagation step run on each policy change

The diagrams

The path of a change under the standard A change starts from a backlog item and an approved spec. An executor agent writes it and a reviewer agent from a different model family reports on it. A deterministic gate, a script that runs in the pre-commit hook and in CI, blocks on a secret, on a broken repository rule and, in the code projects, on a spec that fails validation. In the pull request a bounded reviewer agent and then a person review it before the merge. A separate reminder that a change has no spec only advises. SEPARATION OF DUTIES CODE PROJECTS Backlog item + spec approved before code is written Execute executor agent model family A Review reviewer agent model family B reports, does not fix PRE-COMMIT HOOK + CI Deterministic gate a script, never a model secret scan: blocks specs (code projects): blocks repository rules PULL REQUEST Bounded reviewer an AI first reviewer writes no business code then a person Merge “No spec” reminder advises, never blocks LEGEND Blocks Reviews Advises A change moves forward
The path of a change: what advises, what reviews, and what blocks.
How a policy change reaches the adopting repositories The playbook holds the policies and a registry of the repositories that adopted it. On each policy change a propagation step reads the registry and updates the mirrored documents of every adopting repository. The documents each project owns are left alone. Three repositories are drawn out of ten. The playbook policies · specs · the gate the standard, in its own repository Registry the repositories that adopted it, and which documents mirror it Propagation step run on each policy change POLICY CHANGE READS ADOPTING REPOSITORY Mirrored documents updated Project-owned documents left alone ADOPTING REPOSITORY Mirrored documents updated Project-owned documents left alone ADOPTING REPOSITORY Mirrored documents updated Project-owned documents left alone UPDATE Ten repositories as of November 2026; three are drawn. LEGEND Pushed on a policy change Reads Never overwritten
How a policy change reaches the repositories that adopted the standard.

As presented

Two slides from a presentation of this work, shown as they were presented. They are the conceptual version; the diagrams and the table above are the precise one. Where they differ, the page is right: spec-first applies to the code projects, and review comes after execution.

A slide titled Policy-driven defensive mesh flow. A commit passes a spec-first gate and a secret scan, then a bounded AI agent that reviews without writing code, then an executor model and a reviewer model kept apart, and a deterministic gate, before reaching three synced repositories. A line above shows policy propagating to each stage.
The whole flow in one picture: gates, a bounded reviewer agent, the executor and the reviewer kept apart, and propagation to the adopting repositories. The command under the last gate is illustrative, and the slide draws the reviewer agent before the executor.
A slide titled Gathering requirements: executable specs and fail-closed defaults. On the left, an interview transcript becomes a specification with acceptance scenarios, checked by a strict validation gate. On the right, raw top-k retrieval reaches the generators only through a grounding-refusal gate; the ungated path is crossed out.
From Jobflow, one of the code projects under the standard: requirements written as executable specs, and a retrieval gate that fails closed. The confidence tags in the lower-left box are designed, not shipped.

Two are client engagements, with the clients not named. One is my own practice. The work and the measurements are real.

Rung 03, delivered

A US business-valuation platform, moved and modernised.

Eight web applications and their SQL databases off legacy hosting onto Azure, with failover, backup, monitoring and Zero Trust access added on the way. Then the delivery pipeline, so shipping stopped being an event.

lower running cost
~50%
minutes of downtime
0
start to cutover
5 wks
to release, down from two weeks
2 days

The move

A move and a modernisation, in one pass.

  1. 01Eight web applications and their SQL databases moved off legacy hosting.
  2. 02Failover, backup, monitoring and Zero Trust access added on the way.
  3. 03A delivery pipeline built so shipping stopped being an event.
  4. 04Release time down from two weeks to two days.
  5. 05Cutover in five weeks with no downtime.
For technical evaluators: show the stack

Platform

Azure App Services · Azure SQL with TDE · Blob Storage · Entra ID · Zero Trust network design · CI/CD pipelines · observability and cost governance

AI

Azure AI Foundry · Semantic Kernel · agentic RAG · multi-agent orchestration with guardrails and evaluation. Lead AI and agentic architect on the platform above.

Rung 03, running since 2012

One system has run a retailer's whole business since 2012.

A family-owned tools and machinery retailer in Argentina runs catalog and prices, stock, quotes, invoices, purchase orders and rentals on one system. Email and the website are part of the same engagement. It has grown with the business rather than lagging behind it.

in production
14 yrs
customer accounts
~7,300
sales in US dollars, 2025 vs 2013
~5×
catalog items
~25,000

The system

One system for the whole business.

  1. 01One system has run the whole business since 2012.
  2. 02Catalog by supplier, with cost, list and cash prices, and bulk price updates.
  3. 03Stock with minimums; quotes; invoices that deduct stock; purchase orders that add stock; rental contracts; sales and stock reports.
  4. 04Email and the website are part of the same engagement.
  5. 05In 2026, about 15,500 price-list items were loaded in one import.
For technical evaluators: show the stack

Platform

A .NET CRM on SQL Server, kept current since 2012 · bulk price-list imports · Azure static website · hosted business email

The growth

Every year since 2012, still counting.

Running totals and sales growth from the client's live system, as of November 2026. The totals count records that still exist today.

  • Catalog items, customer accounts and quotes are running totals: records that still exist, not records created that year.
  • Sales are shown as a multiple of 2013, in US dollars. The client's revenue stays private.
  • 2026 runs to 12 November, so its values cover part of the year and are marked as partial.
  • 2026 includes a 15,500-item price-list import, which is why the catalog line jumps.

Catalog items

0 – 25,000

Customer accounts

0 – 8,000

Quotes

0 – 8,000

Sales in US dollars, multiple of 2013

0 – 6.0×

Every year and every value
YearCatalog itemsCustomer accountsQuotesSales in US dollars, multiple of 2013
201254577410no value
20131,1751,219121.0×
20141,5431,595121.9×
20151,8472,025123.0×
20162,3442,3202253.5×
20172,7802,6436003.8×
20183,1192,9211,1282.6×
20193,4563,1431,4521.6×
20203,7643,4241,9771.9×
20214,2863,6692,6901.9×
20225,0153,9473,3752.0×
20236,0444,4384,4581.6×
20246,8665,1845,4353.2×
20258,3396,1696,3204.9×
202624,968 (partial year)7,319 (partial year)7,300 (partial year)5.0× (partial year)
TBDMore case studies

The next one is reserved.

No invented clients, no logo wall, no placeholder metrics. When there is a fourth case study worth publishing, it goes here.

04TBD

Case study 04

Reserved. No placeholder metrics, no invented client, no logo wall.

Start here

Rung one is forty minutes away.

Paid, fixed fee, and I'll say so plainly if you don't need me. Available for remote engagements with US and EU clients.

Córdoba, Argentina — US time zones